Skip to main content
AI & Machine Learning

US Issues Revised AI Executive Order With Weaker Oversight

Government conference room with large table and several chairs, symbolizing AI policy discussion.

"The United States continues to lead the world in Artificial Intelligence (AI) because of the enormous talent and innovation of our AI industry, and because we refuse to stifle this innovation with overly burdensome regulation," the executive order declares — a line that frames the administration's effort to square national-security concerns with a hands-off regulatory stance.

White House: scaled-back voluntary framework and stated priorities

The Trump administration on Tuesday issued a shortened, revised executive order focused on artificial intelligence that abandons earlier, more expansive draft language and adopts a significantly pared-back approach. The directive keeps a largely voluntary framework under which companies may engage with the federal government to test new models before release, but it explicitly states that "nothing in the program will be construed as mandatory or part of a federal licensing or permitting regime." The order is brief — consisting of less than 1,200 words — and frames the administration's dual aims: address national-security risks from rapid model releases while avoiding "overly burdensome regulation" of U.S. businesses.

Frontier-model access: "up to" 30 days, industry influence on scope

A central operational change from earlier drafts is the timeline for government access to so-called frontier models. Where prior language envisioned 90 days of access, the revised order limits voluntary access to "up to" 30 days. The order also gives AI companies "significant influence to help define what models would and would not be covered" under the testing program. It further requires that all federal testing and access be subject to "confidentiality, cybersecurity, insider-risk, and intellectual-property protection, use, and nondisclosure requirements."

Treasury-led cybersecurity clearinghouse and classified benchmarks

The order assigns the Department of the Treasury to head a new interagency cybersecurity clearinghouse on AI. That clearinghouse will be a voluntary forum where "the private sector, critical infrastructure operators and federal agencies" coordinate scanning for software vulnerabilities, discovery and validation, and remediation activities such as patching. Treasury, the Cybersecurity and Infrastructure Security Agency (CISA), the National Security Agency (NSA), the Office of the National Cyber Director, and other agencies are named as responsible for developing classified benchmarks to identify or flag the kind of advanced cyber and hacking capabilities that agencies want to test.

Responses from David Sacks, Senator Mark Warner, and Samir Jain

Tech investor and adviser David Sacks praised the changes on X, calling the revised executive order and the cut from 90 days to "up to 30 days" "a game changer" because, he said, it "would allow frontier labs to comply without delaying new model releases." Sacks also wrote that discussions with the White House "indicate that not all new model releases would be subject to even that level of scrutiny" and that the EO is "intended to apply only to models that represent a meaningful step-change in cyber capabilities (eg Mythos), not to incremental version numbers of existing models."

Senator Mark Warner (D-Va.) said the order would help the White House "begin to grapple" with threats from new frontier models and their hacking capabilities and praised the provision putting the NSA in charge of classified testing. Yet he criticized the administration for reversing course after dismantling prior cybersecurity initiatives, writing that the administration has "belatedly discovered the need to redo something it hastily dismantled in its first year." Warner added that the course correction "can begin to grapple with widespread impacts that new frontier models will have on our critical infrastructure" but warned he will be "watchful" for signs the administration may politicize testing partnerships to "pressure U.S. firms into making changes to their products or Terms of Service to suit partisan or legally questionable objectives of the president and his allies."

Samir Jain of the Center for Democracy and Technology said the order "attempts to avoid the deeply concerning implications of a mandatory licensing regime for release of new models." Jain added that "Testing and benchmarking programs are important to promote cybersecurity and address other risks," while cautioning the EO should not become a mechanism the administration uses "to punish companies for political or other arbitrary reasons."

How technologists, policymakers, and critical infrastructure operators are likely to respond

  • Technologists and security teams: Will watch how "up to" 30 days is operationalized and whether companies retain decisive control over which models enter the voluntary testing program, while evaluating the confidentiality and IP protections the order promises.
  • Policymakers and regulators: Face immediate questions about whether the voluntary, industry-influenced approach will produce meaningful, standardized benchmarks; as American Enterprise Institute fellow Ryan Fedasiuk wrote, "On frontier capability access, vulnerability discovery for critical infrastructure, and sharing with trusted partners, many questions remain."
  • Critical infrastructure operators: Are named participants in the Treasury-led clearinghouse and will be directly affected by how vulnerability scanning, validation, and remediation activities are coordinated under the voluntary regime.

The revised executive order narrows the federal footprint: shorter voluntary access windows, explicit non‑licensing language, and high-level protections for company IP. But the practical weight of the program will be set by implementation — which models qualify as "frontier," how classified benchmarks are developed by NSA and colleagues, and whether the voluntary clearinghouse achieves meaningful vulnerability discovery and remediation. Those specifics, and the watchdoging of potential politicization named by Senator Warner and others, will determine whether the order balances security needs with the administration's stated aim of avoiding burdens on American AI firms.

Source: CyberScoop — Trump administration releases scaled-back AI executive order